Models have for long been used as sketches of what a system does or its specification. Even as sketches, the quality of models is important since high-quality models ease communication between stakeholders. Nowadays, industry tends to use models more and more for tasks other than just describing the system, for example simulation, generating source code and performing various analyses at the model level. Model-driven engineering and domain-specific modelling are approaches that alter the software development process by focusing on developing models that are translated into executable software systems. More attention is therefore paid to the quality of models along with the growing importance of modelling in software development. Quality of models is dependent on several factors such as the quality of modeling languages, modeling tools, modeling processes, and the knowledge and experience of modelers. These factors therefore need attention when discussing the quality of models.The purpose of this keynote at the QUATIC conference is to discuss what model quality means, when and why it matters, and whether we can measure it somehow. We also discuss approaches to improve the quality of models. Model quality can be decomposed in various quality characteristics and these characteristics are used in the talk to discuss which aspect of quality is addressed by which approach. Since model-driven engineering is heavily dependent on tool-support, improving model quality can also be achieved by using tools to assess models for characteristics such as correctness and consistency. Research on the quality of models can promote adoption of model-driven engineering for complex software systems while there are still many questions to answer.